Oracle Adds AI Filters and Data Context Tools to Analytics Cloud

Oracle’s September 2026 Analytics Cloud update promises plainer-language ways to create and distribute analysis. The evidence shows that workbook authors can turn requests into expression filters, while AI-generated dataset descriptions remain material for authors to review and refine; sharing features also depend on administrator configuration.

From plain language to workbook logic

Expression Assistant lets workbook authors speak or type a request for an expression filter. It generates an expression from fields in the selected data source, and authors can apply the result or refine it with a follow-up request. The feature works in workbook and visualization filters, while administrators select the registered language model behind it.

The same release adds AI descriptions for datasets and individual columns. Oracle says this semantic metadata can give AI agents business context that column names may omit, such as differing forecast types, fiscal-calendar formats, or units of measure. Authors can generate a description, but Oracle describes the output as a starting point to check and improve.

Oracle demonstrates using natural language to create expression filters.

Analytics reaches beyond the author

Auto Insights can now be enabled for people viewing a workbook in read-only mode. They can review suggested visualizations, influence the types of suggestions they receive, and export or share insights. Oracle lists examples including trend, seasonality, and growth-contribution views that authors can add to a workbook canvas.

Oracle also adds controlled public links for workbooks, analyses, and dashboards. Administrators must enable public access for an instance and authorize the people or groups allowed to publish. After that setup, recipients can open approved content without signing in to Analytics Cloud.

Developers can embed a chosen workbook canvas in an external web application using sample code, subject to supported authentication and access requirements. Authors retain control through presentation settings over features exposed in the embed, including headers, toolbars, refresh, notes, export, and the Insights panel.

Additional changes around the workbook

• Pivot-table authors can replace null measure values with custom text using Missing Value Text.
• Dataset authors can use descriptor ID relationships so people see friendly descriptors while Analytics Cloud uses corresponding IDs for query processing.
• Oracle’s release documentation also lists preview support for geometry columns in datasets and for ANSI SQL queries through the Analytics Cloud MCP Server.